Part 2 is the matter-of-fact follow-up to Part 1. Spent today inside the actual creative library — the visuals, the copy, the formats, the videos. Watching what's running, what's spending, and how the pieces fit together.
Headline read: there's a real creative engine running on this account with several distinct patterns that are working. It's not one homogeneous winning ad — it's multiple ads doing different jobs in the funnel. That's a healthy sign and gives us obvious places to lean in.
This document is what I see, why I think it's working, and where I'd push next. Section 4 is the centerpiece — five creative directions specific enough to brief a producer. Three quick questions at the end where your context shapes how aggressively to push.
Performance: $1,727 spend / 25 conversions / 0.86 ROAS in 1001 JR Scaling Campaign 1 (last 30d). Also $498 / 10 / 1.50 ROAS standalone in Persist | Testing. Biggest active spender on the account and a winner across two campaigns.
What you see: Listicle-style video with text overlay "Workout Program Not Cutting It?" and a big "5" up front. Direct trial CTA framing — the hook promises a payoff, the body delivers it, the CTA closes.
Hook category: Listicle / Trial CTA
Performance: $686 / 13 conversions / 1.87 ROAS in Persist | Testing. Also running in 1002 JR ABO at $75 / 2 / 6.62 ROAS (1-day sample, small but extreme). Best front-end ROAS on the account.
What you see: Marcus shirtless, kneeling, mid kettlebell curl, orange headband, modern home gym with wood-paneled wall, Assault bike in background. Bold white "SUMMER IS COMING" text. Red/orange "SAVE $76" badge underneath. Aesthetic proof + financial trigger in one frame.
Hook category: Seasonal urgency × Financial save
Performance: $78 spend / 2 conversions in 1002 JR ABO (1-day sample). Early signal but the format is doing identifiable work.
What you see: 2x2 transformation grid. Top half "then" — handstand walk + Olympic snatch in CrossFit-style attire (red shorts, neon knee sleeves, blue institutional gym). Bottom half "now" — controlled DB curl + strict pull-up in modern wood-paneled gym, more refined physique work. Visual narrative does the talking.
Hook category: Transformation × counter-position (away from CrossFit)
Performance: $74 in 1002 JR ABO (early). Separately running as Org variant in Persist | Testing where it had $275 / 6 / 0.91 ROAS. Solid cold-traffic format proven in two places.
What you see: 6-second loop. Not Marcus — appears to be one of the Persist Platinum coaches (short brown hair, blue shirt, lean build). Mid-action pull-ups in a commercial gym (Atlantis-brand machines visible). Static text overlay throughout: "5 Signs You're Training Too Much!" in friendly handwritten-style white font.
Hook category: Pain diagnostic listicle + Coach-led UGC
Performance: $80 in 1002 JR ABO (early). Different doorway from the rest of the lineup — pulls the analytical buyer.
What you see: Dark gray background with 7 light-gray rows showing the weekly schedule. Mon: Lift + Walk. Tues: Walk/Cardio/Recovery. Wed: Lift + Walk. Thurs: same recovery. Fri: Lift + Walk. Sat: recovery. Sun: Walk. Day labels above each row. No founder, no body shot. Clean, minimal, intentionally analytical.
Hook category: Methodology demonstration
A healthy account runs multiple creative roles, not one. The five ads above aren't all doing the same job — they're hitting different people at different stages with different psychology. Mapping the roles below.
The high-spend ads still running in Persist | Testing (NeverGotToLegPress, MovesOver40_Carousel) are doing top-of-funnel prospecting — low frequency, large reach, lower first-purchase ROAS. They're feeding the system new audiences cheaply, not closing. That's expected behavior for prospecting work and explains the lower ROAS — those ads are inputs to the funnel, not the closer.
They're doing two things in one frame, not just one. SummerisComing = aesthetic proof + financial trigger. Thenvnowxfit = identity hook + transformation proof. 5signs = pain callout + curiosity gap. Single-purpose creative is fine and tests well; dual-purpose creative is what scales. The biggest spenders are doing two jobs at once.
"Get Built, Not Burnt." "3/60/10." "Look Good. Move Well." These show up across every winning body copy regardless of format — static, video, infographic, carousel. The reader gets the same brand whether they see a kettlebell shot or a weekly schedule grid. That consistency is doing real work; an account with drift in voice between formats wouldn't compound trust the same way.
Every winning ad ends with "free trial" / "no risk" / "Start Free." Reduces the cost barrier in the conversion math. The 14-day free + 30-day money-back combo from the brand level is being deployed in the creative consistently — not just on the LP. That's smart. Cold-traffic friction comes down when the ad itself, not just the destination, is telling them there's no risk.
SummerisComing is Marcus. Thenvnowxfit is Marcus (looks like). 5signs is a coach (not Marcus). Both convert. That's a structural advantage — it means the brand is the lever, not just the founder. The Persist Platinum coach roster is a creative production multiplier that's already partially in use. Lots of room to do more here without putting all the production weight on Marcus's calendar.
Each of these takes a pattern that's already working and produces variations that hit a different sub-segment, presenter, or emotional angle. Specific enough to brief a producer; not a 30-day calendar.
The 5signs format works (coach + bold listicle text overlay + scroll-stop visual). Produce 4-6 more variations with the other Persist Platinum coaches — Dallas Rogers, Rachel Clark, Adam Fetter, Dominic Scalzo, Erin Dastrup, Alex Critcher, John McGuinness (per the public coaching page). Each carrying their natural angle.
Examples:
Same scroll-stop format, different presenter, different sub-segment hook each time. Production cost stays low because the format is simple — coach in their gym, talking to camera or doing a movement, big text overlay.
The brand was founded on a 6× CrossFit Games athlete walking away from competition because he was beat up. None of the active ads lead with that origin story or the credential. Specific concept: direct-to-camera Marcus saying "I was the 12th-fittest man in the world. Here's why I stopped." Authority + curiosity + counter-position in one frame.
This is a hook the brand has earned and isn't using. Could carry significant creative life — the credential is concrete, the burnout origin is honest, and the counter-position is the same one the brand already runs (Persist vs CrossFit). Just routed through Marcus's own story instead of through transformation grids.
The "then/now" structure works because it speaks directly to people leaving an old training paradigm. Variations:
Same format (2x2 grid, "then" / "now" labels), different "then" narrative for each segment. Each grid speaks to a different sub-segment of the avatar. The first version pulled ex-CrossFit; the others pull adjacent populations the brand should be reaching.
SummerisComing's framing — aesthetic founder shot + bold financial overlay — is doing two jobs in one frame. Variations on the same pattern:
Same structure, different seasonal trigger. Could rotate one of these per month to keep urgency fresh and avoid the audience seeing the same aesthetic+save pairing on repeat.
WeeklyScheduleBnB hits a different person than the aesthetic ads — the buyer who wants to see the system before they buy. More variations of this style:
Text-heavy, no founder, dark-bg infographic style. Carries a different doorway into the brand — the analytical-buyer doorway is currently underserved relative to the aesthetic and identity doorways.
1of10forfatloss_Org_PersistWhat's happening: The ad's hook implies a 10-item listicle ("1 of 10 for fat loss") but the body delivers a counter-position argument about metcons not building lean muscle. Algorithm penalizes this kind of hook/body mismatch — it reads as incoherent and gets throttled.
Fix: Rewrite the body to actually deliver a 10-item list ("10 things to do for fat loss without grinding metcons") OR retitle the hook to match the existing body. Either works. Doesn't require new visual production.
What's happening: The "Running and metcons burn calories in the moment. But they don't build lean muscle mass..." body appears verbatim under both Thenvnowxfit_Persist and 1of10forfatloss_Org_Persist. Algorithm sees them as the same ad — it clusters them and one ends up suppressed.
Fix: Write a unique body for each. Thenvnowxfit gets a transformation-narrative body that talks about the visual journey. 1of10forfatloss gets the actual 10-item listicle body. Two ads, two bodies.
1of10forfatloss performanceWhat's happening: $475 spend / 0.52 ROAS in Persist | Testing — decent on the surface but probably under-performing what it could do once the duplicate-body issue (5b) is fixed.
Fix: Resolve the duplicate-body and hook-body issues before any pause/scale decision on this ad. The numbers we're seeing aren't clean — the algorithm has been rate-limiting it because of the structural issues.
Coach identity on 5signs_Org video — who is the presenter (Dom? Adam? Dallas?), and which other coaches are on-camera-ready in the rotation? This shapes how aggressively to lean into the coach-led video format from Section 4(a).
Has the Marcus authority hook ever been tested? Specifically, has any ad ever led with "6× CrossFit Games athlete" or the burnout origin story directly? If yes-and-failed, I'll drop the proposal in 4(b). If no, it's a real opportunity.
Creative production turnaround for new variations — if I want a new "5 Signs" or "Then/Now" variation with a different coach or different "then" narrative, what does that lead time look like (you / in-house team / outside producer)?
I have detailed visual descriptions for each ad in this doc but didn't embed thumbnails — Facebook image CDN URLs are session-bound and need a separate creative-pull step before they can be cached locally. That's on the build list for v1.1 of the audit pipeline. Not a blocker for the read here, but flagging it so you know thumbnails will land in the next round.